██FR█████ █INTELL███████████

SDLC and AI


This content originally appeared on DEV Community and was authored by Lloyd

What is SDLC?

What can I use AI for in SDLC?

Ideally you already know the scope, goals and requirements. However as we know they can be a bit fuzzy or unclear for certain parts. As such it’s worth working through the SDLC with someone to bounce ideas off such as:

  1. GitHub CoPilot if you’re storing everything in the GitHub Repositories which is my personal platform of choice.
  2. M365 CoPilot if you’re storing items in SharePoint, Outlook, Word, Excel, Streams. My personal favorite for converting transcripts from meetings into Action items, Converting Emails into an executive summary depending on who I’m communicating to and what they’re asking.
  3. Amazon Bedrock for most of your AWS hosted solutions you want to use GenAI with.
  4. Azure OpenAI for most of your Azure hosted solutions you want to use GenAI with.

How does it map to SDLC?

Planning

A lot of planning takes place in meetings where we can loose track of the scope, agenda and point of the meeting. I’m always slightly wary of a meeting with no scope, links or proactive / reactive response to something we have to do.
As such it’s best to write up the scope of the meeting, or propose one based on what you think it’s about (1 – GHCP).
If you’re not proactively planning out the roadmap (8 -GHP), then you may not be prepared as things kick off.

Analysis

A lot of analysis can happen in meetings which is why I strongly advocate for transcripts to be converted into documentation, action items and outstanding questions (2 – M365CP)

Design

Designing the project architecture is probably the least automatable of these. However Templating these so the format that is predictable and feeds into proactively meeting requirements down the line in SDLC is well worth it (1 – GHCP). Remember Markdown is probably the best format for these.

Coding

Lots to cover, I’ll just say the design artifacts should feed into this heavily (1 – GHCP).

Testing

Lots to cover again, I’ll just say the design artifacts should feed into this heavily (1 – GHCP).

Deployment

Finally, we’re at the Platform Engineering phase of DevOps where the developers and engineers are treated like customers rather than force fed some horrible take on how DevOps should work without allowing us to improve it.
If you don’t know what it is, start here.
If you’re platform is written well the previous phases should feed into this one nicely (1,2,3,4,8).

Maintenance

I haven’t forgotten about maintenance, but it needs to be kept in mind that someone has to maintain this. As we shift left, the engineer writing the code is hopefully supporting the code so they have a major stake in making it maintainable.
Again, if you’re platform is written well the previous phases should feed into this one nicely (1,2,3,4,8).

What did I use recently?

For the .NET Meetup I used Azure AI as I wanted to use:

  1. .NET.
  2. Predictable responses.
  3. Gamify my Agile board with some theme that’s fun for the team (my family).
  4. GitHub Projects for a Career Roadmap for myself that I can template for others.

More on this later…


This content originally appeared on DEV Community and was authored by Lloyd